0360.3016/84 $03.00 + .

xl
(opynghr Q 1984 Pcrgamon Press f.,d

??Original Contribution

THERMAL DOSE DETERMINATION IN CANCER THERAPY

STEPHEN A. SAPARETO, PH.D.

Division of Medical Oncology, Department of Internal Medicine, Wayne State University School of Medicine,
P.O. Box 02188, Detroit, Michigan 48201

WILLIAM C. DEWEY, PH.D.

Radiation Oncology Research Laboratories, University of California CED 200, San Francisco, California 94 143

With the rapid development of clinical hyperthermia for the treatment of cancer either alone or in conjunction
with other modalities, a means of measuring a thermal dose in terms which are clinically relevant to the biological
effect is needed. A comparison of published data empirically suggests a basic relationship that may be used to
calculate a thermal dose. From a knowledge of the temperature during treatment as a function of time combined
with a mathemadal description of the time-temperature relationship, an estimate of the actual treatment calculated
as an exposure time at some reference temperature can be determined. This could be of great benefit in providing
a real-time accumuhted dose during actual patient treatment. For the purpose of this study, a reference temperature
of 43C has been arbitrarily chosen to convert ail thermal exposures to equivalent-minutes at this temperature.
This dose calculation can be compared to an integrated calculation of the degree-minutes to determine its
prognostic ability. The time-temperature relationship upon which this equivalent dose calculation is based does
not predict, nor does it require, that different tissues have the same sensitivity to heat. A computer program
written in FORTRAN is Included for performing calculations of both equivalent-minutes (4,) and degree-minutes
(t,&. Means are provided to alter the reference temperature, the Arrhenius break temperature and the time-
temperature relationship both above and below the break temperature. In addition, the effect of factors such
as step-down heating, thermotolerance, and physiological conditions on thermal dose caiculntions are discussed.
The equations and methods described in this report are not intended to represent the only approach for thermal
dose estimation; instead, they are Intended to provide a simple but effective means for such calculations for clinical
use and to stimulate efforts to evaluate data in terms of therapeutically useful thermal units.

Hyperthermia, Thermal dosimetry, Arrhenius analysis, Equivalent-minutes.

INTRODUCIION situations, the time required to achieve a predetermined

With the rapid development of clinical hyperthermia for
the treatment of cancer either alone or in conjunction
with other modalities, a means of measuring a thermal
dose in terms which are clinically relevant to the biological
effect is needed. Clinical trials for local hyperthermia
treatment have been rep~rted;~*~~*~~ however, there has
been little consistency in protocols. The temperatures
used have varied from 42 to 50 Celsius, and the duration
of treatment and the frequency and number of multiple
fractions have also varied. The difficulties in assessing
and comparing these treatments are obvious.
A plethora of data on the effect of heat on both cells
in culture and in tumors is available, A comparison of
these data empirically suggests a basic relationship that
may be used to calculate a thermal dose. In clinical
hyperthermic exposure temperature can be a significant
part of the total treatment time. Furthermore, in some
instances, it may not be possible to achieve a prechosen
temperature because of such limitations as patient dis-
comfort, insufficient power, tumor location, high tumor
blood flow, or a combination of these factors. Therefore,
it is essential to determine some sort of comparative dose
estimate for the actual treatment given. From a knowledge
of the temperature during treatment as a function of time
combined with a mathematical description of the time-
temperature relationship for thermal inactivation or
damage, an estimate of the actual treatment calculated
as an exposure time at some reference temperature can
be determined. For the purpose of this report, a reference
temperature of 43C has been arbitrarily chosen.

This work was supported in part by grants CA22453 and Reprint requests to: S. A. Sapareto, Ph.D.
CA3 18 13 from the National Cancer Institute, DHHS. Accepted for publication 6 February 1984

787
788 Radiation Oncology 0 Biology 0 Physics June 1984. Volume 10. Number 6

The importance of this technique can be seen in its where R can be calculated as a function of AH, the ac-
application to determining the dose accumulated in real tivation energy (cal/mol), and absolute temperature (K)
time during actual exposure, in order that hyperthermic from an Arrhenius plot by:
treatment could be adjusted for temperature variations
during the actual treatment. For example, in a case where (2)
the measured temperature exceeded the proposed treat-
The constant 2 is an approximation for the universal
ment temperature, the actual treatment could be short-
gas constant ( 1.98 Cal/OK-mol) and the numerator of the
ened during the exposure to correct for the extra damage
exponent contains a constant of 1K which cancels the
occurring during the period of excessive temperature.
unit of K in the denominator, thus providing the correct
Other methods of calculating thermal dose have also
been proposed..8 These methods are based also on a dimensions. Although R is a function of temperature, for
thermodynamic or Arrhenius-type relationship which the range of interest (37-46C) assuming that R is con-
has been empirically determined. The equations and stant will give an error of less than 2%. The selection of
methods described in this report are not intended to rep- R values can significantly affect the calculated dose; how-
ever, this occurs only when the actual temperature is
resent the only approach for thermal dose calculation;
different from the reference temperature. A change in R
instead, they are intended to provide a simple but effective
of 10% will cause approximately a 10% error in the cal-
means for such estimations for clinical use and to stim-
culated dose per degree difference between the actual
ulate efforts in evaluating clinical results in terms which
temperature and the reference temperature. Determi-
are practical and prognostically relevant.
nations of R have been reported for a number of biological
systems and endpoints (see review in reference 5). The
values reported range from 0.4 to 0.8 above 43C with
THEORETICAL MODEL
0.5 being the most common value. There have been fewer
studies below 43C; however, in general the R value is
Equivalent-minutes calculation
approximately a factor of 2 smaller than that above
The relationship between temperature and exposure
43C.3.5.26,29Thus, we have assumed R = 0.5 for tem-
time during hyperthermic treatments has been reported
peratures greater than 43.OC and R = 0.25 for tem-
for a variety of biological systems.5.6,sThe evidence clearly
peratures below 43.OC.
indicates that, for both in vitro and in vivo systems, an
The choice of the break temperature at 43C has
exponential relationship exists between temperature and
also been arbitrarily chosen as a best estimate from all
exposure time. In most systems, this relationship can be
available data. However, there is evidence that this tem-
simply stated: a one degree increase in temperature re-
perature may vary2,* and that it may be a relative tem-
quires a two-fold decrease in time for the same effect
perature related to the normal physiological temperature
above 43C and a three-to-four fold decrease in time for
of the tissue rather than an absolute value.
an iso-effect below 43C. Investigators have attempted
For the simple case of equating a time at one tem-
to describe this relationship mathematically;~5~8~9how-
perature with an equivalent time for the same effect at
ever, these analyses have been directed primarily toward
another temperature, a nomogram can be drawn (Figure
the mechanisms of heat killing, and little effort has been
I). This nomogram has a reference temperature of 43C
put into the clinical application of this mathematical ap
and can be used in two ways. First, if a preselected treat-
preach.
ment at the reference temperature is chosen, the appro-
As stated above, the overwhelming majority of bio-
priate time at any other temperature to achieve the same
logical systems exhibit the same exponential relationship
effect can be calculated. Conversely, if a treatment at
between time and temperature for a given &effect. For
some temperature is given, this may be equated with an
cells in culture, this is most clearly observed in an Ar-
equivalent treatment time at 43C. Therefore, compar-
rhenius plot of the logarithm of the reciprocal of Do versus
isons between treatments at different temperatures can
the reciprocal of temperature.3,5.5.35 This plot is linear
be made by converting each treatment to an equivalent
and of approximately the same slope above 43C for
time at 43C.
almost all cell lines.3.5 Of course, this is not to say that
An important extension of the use of this relationship
all cells exhibit the same sensitivity to heat, as evidenced
is the determination of the equivalent time at 43C for
by the vertical displacement between lines on an Arrhenius
a complex temperature history. As shown in Figure 2, a
plot, 5*35but only that for many cell lines, a consistent
typical temperature profile for a treatment would consist
relationship describes how the slope of the survival curve
of three components: 1) an initial warm-up period (often
changes with temperature.
exponentially approaching the treatment temperature),
This same relationship is also seen for many in vivo
followed by 2) a period of more or less constant tem-
systems26 and mathematically has been described as a
perature, and ended by 3) a cool-down period (also typ-
relationship between time (t) and temperature (T) by:
ically exponential). Ideally, the least complicated tem-
perature history would have instantaneous temperature
t, t2R(TI-T2)
transitions and a period of constant temperature mainte-
zz
(1)
 Thermal dose 0 S. A. SAPARETO AND W. C. DEWEY 78Y

IO 100
46

---.--------- . . . . .._.. ____i__

____...-----
4*. _______..------- 50
________.-.---
_____._....-----
43

50
42

loo IO
41

T
-- 5

500 1u_ t&WV

IVWSE~R=05 T>43 PC
~ R=025 T<43O=C
1000 ---I

t f43

Fig. 1. A nomogram relating time at any temperature to an equivalent time at 43C (t4& Two examples of use
are shown by dashed lines; a 30 minute treatment at 44C is equivalent to 60 minutes at 43C which is also
equivalent to IS minutes at 45C.

nance as also shown in Figure 2. For ease of comparison, By mathematically describing the change in temper-
this is the temperature profile to which all complex profiles ature as a function of time, it is possible to calculate the
should be reduced. equivalent time at any chosen reference temperature. A

I I I I I I I I I I I 1 1

47

37
I 1 I I I I I I I 1 L I I
0 20 40 60 80 100 120 140 160 I80 200 220 240

TIME (min.)
Fig. 2. Theoretically generated temperature profiles as a function of time. The ideal temperature profile is shown
by the dashed line and represents 120 minutes at 43C. The three additional profiles (A-C) are all equivalent to
120 minutes at 43C by equivalent-minute calculation. As can be seen, the areas under each curve are nof equal.
790 Radiation Oncology ??Biology0 Physics June 1984, Volume 10, Number 6

common case is that of an exponential change in tem-
perature of the form:
T(t) = Tr - ATem
where Tr is the final temperature, AT is the difference
between the final and initial temperatures, and X = 0.693/
t,,z, with tllZ being the half-time for the temperature
change.
Henle and Roti Roti* have evaluated the equivalent
treatment time for the type of exponential warm-up de-
scribed in equation 3. To determine the total exposure
time, it is necessary to integrate the time-temperature
relationship as a function of time. Since, in this case, the
equation to be integrated is an indefinite integral of the
form e/x, the authors used an approximation by series
expansion limited to the first 10 terms.
A more general approach to a changing temperature
exposure would be to calculate the accumulated exposure
by numerical integration using a computer. This provides
a method for calculating the accumulated dose at a ref-
erence temperature under a variety of heating profiles,
including temperature histories that cannot be easily de-
scribed mathematically.
For sufficiently small At, the equation can be described
mathematically as:
1=linal
t 43 = 2 R43-i,At
1=0
where t.,) is the equivalent time at 43C, 7 is the average
temperature during time At, R = 0.5 above 43C and
R = 0.25 below 42C.
The frequency with which temperature measurements
are made (At) can affect the accuracy of the thermal dose
calculation. Obviously, the more frequently the mea-
(3) surements (or the smaller the At), especially during rapid
temperature changes, the more accurate the calculated
dose. On the other hand, in situations where the tem-
perature is changing very slowly, less frequent measure-
ments may provide sufficient accuracy. For the theoretical
curves used to generate the results in this report, At was
less than 0.2% of the total treatment time. Thus, at least
500 temperature measurement points per treatment were
used.
In order to completely describe a thermal dose, several
new terms must be defined. These terms must be easily
identifiable treatment parameters. First, t,,,, is defined as
the time from the start of thermal power input until the
end of power input. Thus, t,,,l includes the warm-up but
does not include the cool-down to normal temperature.
Second, kO, is defined as the correction time to be added
to the total time (ttotal) to account for either the warm-
up or cool-down periods. Hence, warm-up corrections
are negative in value and cool-down corrections are pos-
itive. Conversion of t,,,l or torr to equivalent times at
43C may be made by use of Figure I and would then
be designated as t,olal43or &0rr43,respectively.
The importance of correcting for the warm-up period
is clearly demonstrated in Figure 3. This figure represents
theoretical temperature histories for a 30 minute (t,,,J
(4) treatment at 45C with various exponential warm-up half-
times. Using Figure I, the equivalent total time (tlola,43)
of this treatment with instantaneous warm-up and cool-
down would be 120 minutes. As shown, a 10 second half-
time warm-up causes a negligible reduction of the equiv-
alent treatment time to 117.7 minutes (&0rr43= -2.3

% 1,,= 93.0 min

j t-
=B TIME(MIN)

m-
t,,z 106.5 min - 3- 1,*: 66.7 min
36 ,,), ,Ir1 ,,(, ,,, ,,,, ,1,, 36--,,,, ,,,, ,((, ,,,, ,/,, ,,,,
a 5 LB IS a 25 1 a 5 *m LS a 25 Ia
TIME (Ml,,) TIME (b&N)

Fig. 3. Theoretically generated temperature profiles as a function time for a treatment temperature of 45C. Each
represents an equivalent total time at 43C (t toU143)of 120 minutes with different half-times for warm-up. Half-
times and equivalent-minutes at 43 (4,) are indicated.
 Thermal dose 0 S. A. SAPARETO AND W. C. DEWEY
791

minutes). However, for a four-minute half-time warm-
up, which is a quite likely clinical occurrence, the equiv-
alent treatment is reduced to 56% of the total time (tcorr43
= -53.3 min). Note that a rough estimate from Figure
3 of the actual time at the final temperature (e.g. 15 min
at 45 for tl,2 = 4 min is a reasonable approximation of
the dose when converted to equivalent-minutes (4, = 60
min). Thus, the common clinical practice of determining
the duration of treatment based on the time spent at the
treatment temperature may not be too unreasonable.
An empirically-derived relationship between tcorrd3and
the half-time for warm-up or cool-down, where the half-
time is converted into an equivalent half-time at 43C
(tl,2 &, can be estimated as shown in Figure 4 for half-
times of less than 15% of the total time. As can be seen,
the warm-up correction is an order of magnitude greater
than the cool-down correction for a given transition half-
time. This occurs because during warm-up, the temper-
ature rises slowly through the higher, more toxic tem-
peratures before reaching the treatment temperature,
while during cool-down, the temperature drops rapidly
to the lower, less toxic temperatures.
EXPERIMENTAL DATA
Equivalent-minutes calculation
The applicability of equivalent-minute calculations is
shown in Figure 5 where the in vitro survival of cells
exposed to temperatures from 4 I .5 to 46.5C for various
times are plotted as a function of equivalent-minutes at
43C. The data fit a single curve (r2 = 0.87) with the
exception of prolonged heating durations of 3 hours or
more at temperatures of 42.5C and below. At these tem-
peratures, survival follows the curve initially, but then
deviates from the single curve because of the development
of increased thermal tolerance during the heat treatment.29
Furthermore, as temperature decreases from 42.5 to 4 I .5,
the data appear to deviate from the single curve at earlier
equivalent times, which reflects the fact that maximum
thermotolerance develops at different survival levels,29
because the rate of killing increases with temperature.
The calculation of equivalent-minutes at 43C does,
in fact, partially correct for thermotolerance, since the
break in the Arrhenius plot at 43.OC is most likely
due to the development of thermotolerance during heat-
ing. 22,29However, this correction does not sufficiently ac-
count for thermotolerance (Figure 5) as maximum tol-
erance is reached, i.e., a plateau in the survival curve.29
Since the rate of the development of thermal tolerance
is probably temperature dependent. the assumption of a
constant value for R below 43.OC may not be valid, and
certainly does not apply when maximum thermotolerance
occurs, where R would become infinite. Indeed, these
relationships require further investigation.
Degree-minutes calculation
While the approach described above is based on ex-
tensive empirical data, an appropriate model for com-

-60 +6

0 2 4 6 8 IO 12

t ~43 (mid
Fig. 4. An empiricallyderived graph of warm-up (solid line, lef? ordinate) and cool-down (dashed line, right
ordinate) correction times at 43C as a function of equivalent half-time at 43C. The correction time (fM3) can
be determined by converting the warm-up half-time for the actual treatment temperature to an equivalent half-
time at 43C by the nomogram in Figure 1 (e.g. 10 sec. half-time to reach 45C equals a 40 second half-time to
reach 43C). The straight lines are valid for half-times up to 15% of the total treatment time and each line is
valid over the temperature range indicated in the figure.
792 Radiation Oncology 0 Biology 0 Physics June 1984, Volume 10, Number 6

pendix I. The program, called TEQUIV, was written in

FORTRAN 4 on a PDPl l/23 compute? and should be
08 0 0 41.5 sufficiently documented. Two alternate subroutines have
EA-o_ v 42.0
been provided for data input either directly from an in-
F) A 42.5
t + 43.0 teractive terminal or from a data file. The program allows
7.
L% Cl 43.5 changes of R values and the Arrhenius break temper-
v 44.0 ature; however, default values used in this report are pro-
vided. Further information about this program including
test data to verify its accuracy and a version of the program
written in BASIC can be obtained from the authors. In
addition, a program will be available to provide real-time
accumulated dose with the input of temperature values
during actual treatment.
A
A
--__
COMPLICATIONS AFFECTING
THERMAL DOSE
Step-down heating
A
An initial exposure of cells to temperatures above 43C
causes a modification of the time-temperature relationship
below 43C. If cells are briefly exposed above 43C and
then immediately treated below 43C, the break in the
Arrhenius plot at 43C is eliminated; thus an R value of
0.5 is maintained over the whole temperature range.13.**
t,, (mid
Both Sapareto et al.29 and Li et al.* suggest that the
development of thermotolerance causes the break in the
Fig. 5. The dose survival response for asynchronous Chinese Arrhenius plot and that exposure to temperatures above
hamster ovary cells at various temperatures plotted as a function 43C inhibits or delays the development of thermotol-
of equivalent-minutes at 43C. Error bars have been omitted
for clarity. The data at 4 1.5,42.0, and 42.5 deviate from a single erance, thus allowing more rapid killing when cells are
line, as shown by the dashed lines, due to the development of subsequently exposed to temperatures below 43C. This
thermotolerance. Actual data are taken from reference 29 and phenomenon, of course, would affect the calculation of
replotted. an accumulated dose as described in the model presented
here by increasing the f4) dose accumulated during ex-
parison would be the integral of the product of the tem- posure to temperatures below 43C. However, by deter-
perature above a certain starting point multiplied by time, mining the amount of high temperature exposure nec-
or the degree-minutes. This degree-minute quantity essary to cause this effect, the R value below 43C should
can be reconverted back into a time as if the whole period be adjusted after the appropriate accumulated exposure
had been at a constant temperature by dividing by the necessary to reflect that this change had occurred. For
difference between the starting temperature and reference example, Figure 3, curve B, represents a step-down heating
temperature. Again, using 43C as a reference, this is situation. Following these arguments, this curve would
expressed as: represent a dose greater than 120 minutes at 43. Further
t=hlXd
research is essential to develop and test these possibilities.
2 (-i- T,)At
t=o Multiple dose therapy
tdm43 =
A simple calculation ofequivalent time at 43C cannot
43 - To
be accomplished for multiple heat doses because of the
where &,43 is the degree-minutes converted to a time at development of thermal tolerance between treatments.
43C, and To is the starting temperature which has been As has been clearly demonstrated for a variety of cells,
arbritrarily chosen to be 4 1C since there is little evidence whether normal or malignant, previous exposure to el-
of cell killing below this temperature. evated temperature produces resistance to further thermal
damage, for periods of up to 6 day~.~~~~~~*~The effect of
Computer programs the development of thermal tolerance on the Arrhenius
A program performing calculations of both equivalent relationship is to cause both a shift in the break tem-
minutes (t& and degree-minutes (tdrn43)is listed in Ap perature to higher temperatures and a displacement of

* Digital Equipment Corporation, Maynard, MA.

 Thermal dose 0 S. A. SAPARETO AND W. C. DEWEY 793

the linear relationship toward slower rates of killing (i.e.,
higher D,, values). 2.20However, the slope of the Arrhenius
relationship does not appear to change.2,20 This is dem-
onstrated for in vivo normal tissue in Figure 6.
Based on these observations, the fundamental rela-
tionship between time and temperature used in the cal-
culation of equivalent-minutes is still valid for thermo-
tolerant cells. However, the f4) dose calculated must be
reduced to account for the degree of thermotolerance
present. A measure of this degree of thermotolerance has
been proposed by Henle and Leeper as a ratio of Do
values from survival curves for Chinese hamster ovary
cells, with or without thermotolerance, and was termed
the therrnotolerance ratio (TTR). Thus, this phenomenon
appears to be a dose-modifying factor similar to that for
the oxygen effect (OER) seen with cell killing by ionizing
radiation.12
The TTR which can be induced with acute heating
treatments ranges from approximately 2.4 to 4.5.2.417
Note that this is not the same as the ratio of survival
levels. Figure 620 also shows a ratio of approximately 3
between doses in the presence and absence of thermo-
tolerance for the same normal tissue damage. Thus, it
may be possible to determine some dose correction factor
if the rate and degree of thermotolerance development
can be predicted. Further investigation of these dose
modifying concepts are necessary to determine the fea-
sibility of this approach.
An additional complication of thermotolerance is the
development of tolerance during the warm-up period of
the heat treatment. As noted previously, the R value of
0.25 below 43C does partially account for some of this
tolerance. However, slow rates of heating would also cause
the development of thermotolerance to the higher tem-
peratures finally achieved during the same treatment.
Although maximum thermotolerance requires several
hours to develop, 29.32the development probably begins
immediately upon initiation of heating;23 therefore, the
rate of heating to treatment temperature should be kept
as rapid as possible, certainly within one-half hour or
less.
Physiological conditions
Several physiological factors such as pH and nutrients
also modify cellular response to heat;, for example,
lowering the pH, although having little effect on R, in-
creases the cellular inactivation rate. Furthermore, these
factors can affect the development of thermotolerance.0*23
However, again, these phenomena may be similar to those
which affect ionizing radiation response (e.g. OER) and
can be treated as dose-modifying factors. In fact, Goldin
and Leeper have demonstrated a simple relationship
between pH and the maximum TTR achieved by 45C
exposure. These factors change appreciably in tumors,3.33
but should be relatively constant in normal tissues; there-
fore, complications related to these factors should not

ReciprocaI of temperature ( K-)

320 318 316 314 312dO-5

I 1 1 . ,Ouol

.; 400 - \ - 0002

\
\
\
EE 200 - -o004~
._
5 E

L
; 100 - -001 ;
L
e! 8
6 Ncrmcll ears -Oo2 2
5 40-
W
25 20 - -004 a

lo 40
1 421 44
L 46
1 ce

Temperature (C 1
Fig. 6. An Arrhenius-like plot of the relationship between time and temperature to cause 50% necrosis in mouse
ears. Resistant ears received a treatment of 43.5C for 20 min., 24 hours prior to the teSt treatment. Reprinted
with permission from reference 20.
794 Radiation Oncology 0 Biology 0 Physics June 1984, Volume IO, Number 6

pose a serious problem in establishing heat doses for nor-
mal tissues.
Hyperthermia plus radiation
The calculations of thermal dose presented in this report
are based on the killing effects from heat alone. When
heat is combined with radiation, there are two components
to the increase in cell killing: 1) killing from heat alone,
and 2) increased killing from radiation due to the inter-
action of heat combined with radiation. Law et aL2 have
shown that the combination of heat plus radiation ap-
parently alters the Arrhenius relationship, giving an R
value of approximately 0.27 over the temperature range
of 41 to 44.5%. In addition, we have notedm that com-
bining a single dose of radiation with different heat doses
over the range of 41.5 to 45.5C, each adjusted by an
equivalent-minutes calculation to give the same killing
from heat alone, had a maximum rate of inactivation at
42.5C, i.e., about 10 to 15% greater than at 41.5 or
45.5C. This, too, suggests that the R value chosen from
heat killing data may not be the best for heat interacting
with radiation. However, both of these studies, as well as
others,16,24 are of limited use because they do not take
into account the effect of differences during the cell cycle
in sensitivities to heat and radiation3 A study of ho-
mogenous populations of Gl and S phase cells heated
and irradiated without such complications28 suggests that
differences in inactivation rates due to the interaction of
heat and radiation at equivalent thermal doses are not
large (less than 10-l 5%). Nevertheless, in order to truly
understand the complications in calculating thermal dose
for the combination of hypertheirnia and radiation, fur-
ther studies with homogenous populations of cells in
which heat radiosensitization is clearly separated from
heat killing are necessary.
shown in Table 2, for a total time of 30 minutes at 45C
or a &i43 of 120 minutes, the two calculations differ
dramatically both in absolute dose and in relative re-
duction in dose with longer warm-up times. Table 3 per-
forms ,the same comparison for a total time of 480 minutes
at 42C which is also a ttouls3of 120 minutes, and again
the two methods differ greatly in estimating the dose.
Thus, only when the average treatment temperatures differ
from 43C can the two methods be compared in clinical
trials to determine which is a better prognostic indicator
of therapeutic effect.
The time-temperature relationship upon which this
equivalent dose calculation is based does not predict, nor
does it require, that different tissues have the same sen-
sitivity to heat. These calculated doses are valid only as
a means of comparing treatments on the same tissue and
for the same effect. Thus, once the effect of a given treat-
ment is known, such as the maximum tolerated thermal
treatment for a given normal tissue or the minimum
tumor treatment for a given probability of cure, different
treatments at different temperatures can be either pre-
dicted or compared.
The importance of equivalent-minutes as a prognostic
indicator of treatment response has been investigated by
Dewhirst et aL4 in clinical trials of spontaneous domestic
pet tumors. The results clearly indicate that the thermal
dose measured for the thermometer probe in the coolest
part of the tumor is the best predictor of long-term re-
sponse. In addition, the equivalent-minute dose (t43) is a
better prognostic indicator when compared to a degree-
minute dose (&,,43). Furthermore, the dose determined
for the first treatment in multiple dose therapy was better
than the total accumulated dose at predicting long-term
response. This higher correlation with first heat dose is
further evidence that thermotolerance may reduce the

CLINICAL EVALUATION Table 2. Comparison of calculated doses for a treatment of

In order to determine the relevance of either equivalent- 30 minutes at 45C for various warm-up half-times
minutes or degree-minutes for measuring thermal doses
t,,z@ec) t3tmin) tdd3(min)
in clinical treatment, it is essential to determine under
which conditions the models differ. Table 1 compares the 20 115.5 58.4
dose calculations for each of these methods with various 40 111.0 56.7
80 102.0 53.5
warm-up half-times for a ttoti43 dose of 120 minutes. In 160 84.0 47.0
this case, no significant difference between the two meth- 320 51.3 34.6
ods can be seen. However, at a higher temperature, as

Table 1. Comparison of calculated doses for a treatment of Table 3. Comparison of calculated doses for a treatment of
120 minutes at 43C for various warm-up half-times 480 minutes at 42C for various warm-up half-times

h(~4 t3(min) td,43(min) tl Aec) t43(min) tdd3(min)

20 118.7 119.0 20 119.7 239.3

40 117.4 118.0 40 119.4 238.7
80 114.8 116.0 80 118.7 257.5
160 109.6 111.9 160 117.5 235.0
320 99.2 103.9 320 114.9 230.0
 Thermal dose ??S. A.
effectiveness of the later treatments in multiple dose ther-
apy, despite the separation of treatments by 72 hours or
more.
The clinical usefulness of calculating an equivalent
thermal dose depends on several factors. The first is that
the Arrhenius relationship (AH and the break tempera-
ture) be valid for the tissues being treated. This linear
time-temperature relationship above the break temper-
ature has been shown to exist for the vast majority of
both normal and tumor tissues studied.3~5~5~9~20~26~35
Nev-
ertheless, even the possibility of a variation in the Ar-
rhenius relationships for human tumors will not lessen
the usefulness of these dose calculations. In fact, even
complicating factors such as changes in pH, nutrients,
and blood flow may have only minimal consequences
on thermal dose determinations, once their effects on
inactivation rates are quantified.
Tubiana34 has stated emphatically that the goal of any
REFERENCES
1. Atkinson, E.R.: Hyperthetmia dose definition. J. Bioeng.
1: 487-492, 1977.
2. Batter, K.D., Henle, K.J.: Arrhenius analysis of heat survival
curves from normal and thermotolerant CHO cells Radiat.
Rex 78: 251-263, 1979.
3. Connors, W.G.,Gemer, E.W., Miller, R.C., Boone, M.L.M.:
Prospects for hyperthermia in human cancer therapy. Part
II. Implications of biological and physical data for appli-
cations of hyperthennia to man. Radiol. 123: 497-503,
1977.
4. Dewhirst, M.W., Sim, D.A., Sapareto, S.A., Connor, W.G.:
The importance of minimum tumor temperature in de-
termining early and long term responses of spontaneous
pet animal tumors to heat and radiation. Cancer Res. 44:
43-50, 1984.
5. Dewey, W.C., Hopwood, L.E., Sapareto, S.A., Gerweck,
L.E.: Cellular responses to combinations of hyperthermia
and radiation. Radiology 123: 463-479, 1976.
6. Field, S.B.: The response of normal tissues to hyperthetmia
alone or in combination with x-rays. In Cancer Therapy
by Hyperthermia pnd Radiation, Streffer, C. (Ed.). Munich,
Urban and Schwarzenberg. 1978, pp. 37-48.
7. Freeman, M.L., Raaphorst, G.P., Hopwood, L.E., Dewey,
W.C.: The effect of pH on cell lethality induced by hyper-
thermic treatment. Cancer 45: 2291-2300, 1980.
8. Gemer, E.W.: Thermal dose definition and evaluation in
terms of hypetthermia induced cytotoxicity (Abstr.). 7th
Int. Congress of Radiat. Res. Amsterdam, Netherlands,
1983.
9. Gerweck, L.E., Jennings, M., Richards, B.: Intluences of
pH on the response of cells to single and split doses of
hypetthermia. Cancer Res. 40: 40194024, 1980.
10. Goldin, E.M., Leeper, D.B.: The effect of low pH on ther-
motolerance induction using fractionated 45 hyperthermia.
Radiat. Res. 85: 472-479, 198 1.
11. Hahn, G.M.: Metabolic aspects of the role of hyperthermia
in mammalian cell activation and their possible relevance
to cancer treatment. Cancer Res. 34: 3117-3123, 1974.
12. Hall, E.J.: Radiobiology for the Radiologist, 2nd edition.
Hagerstown, Harper and Row, Inc. 1978, pp. 79-92.
13. Henle, K.J.: Sensitization to hyperthermia below 43C in-
SAPARETO AND W. C. DEWEY 795
cancer therapy is a maximum destruction of malignant
cells with minimal or acceptable normal tissue damage;
therefore, the ultimate doses for hyperthermic cancer
treatment will be limited by normal tissue damage. Unlike
tumors, normal tissue is more likely to show a predictable
and generally applicable time-temperature relationship
which will allow therapists to treat to an accurately pre-
determined maximum dose. This dose predictive capa-
bility is desperately needed since thermal damage fre-
quently exhibits a threshold-like response where, once a
certain dose of heat is given, only a slight increase in dose
can cause a tremendous increase in tissue damage. In
order for hyperthermia to achieve its true therapeutic
potential, treatments must approach this threshold level
as closely as possible. Improvements in heat delivery,
temperature measurements, and thermal dose calculations
during the actual treatment period are the only way this
can be achieved.
duced in Chinese hamster ovary cells by step-down heating.
J. Natl. Cancer Inst. 64: 1479-1483, 1980.
14. Henle, K.J., Dethlefson, L.A.: Heat fractionation and ther-
motolerance: A review. Cancer Res. 38: 1845-185 1, 1978.
15. Henle, K.J., Dethlefsen, L.A.: Time-temperature relation-
ships for heat-induced killing of mammalian cells. Ann.
N. Y. Acad. Sci. 335: 234-253, 1980.
16. Henle, K.J., Leeper, D.B.: Interaction of hyperthermia and
radiation in CHO cells: recovery kinetics. Radiat. Res. 66:
505-5 18, 1976.
17. Henle, K.J., Leeper, D.B.: The induction of thermotolerance
in CHO cells by high (45) and low (40) hyperthermia.
Cancer Res. 38: 575-578, 1978.
18. Henle, K.J., Roti Roti, J.L.: Time-temperature conversions
in biological applications of hyperthermia. Rad. Res. 82:
138-145, 1980.
19. Henriques, F.C.: Studies of thermal injury: V. The Pre-
dictability and the significance of thermally induced rate
processes leading to irreversible epidermal injury. Arch. Path.
4: 489-502, 1947.
20. Law, M.P.: Induced thermal resistance in the mouse ear:
The relationship between heating time and temperature.
Int. J. Radiat. Biol. 35: 481-485, 1979.
21. Law, M.P., Ahier, R.G., Field, S.B.: The response of the
mouse ear to heat applied alone or combined with X rays.
Br. J. Radiol. 51: 132-138, 1978.
22. Li, G.C., Cameron, R.B., Sapareto, S.A., Hahn, G.M.:
Reinterpretation of Arrhenius analysis of cell inactivation
by heat. NC1 Monograph 61: 11 l-l 13, 1982.
23. Li, G.C., Hahn, G.M.: A proposed operational model of
thermotolerance based on the effects of nutrients and the
initial treatment temperature. Cancer Res. 40: 4501-4508,
1980.
24. Loshek, D.D., Chr, J.S., Solomonidis, E.: Interaction of
hyperthermia and radiation: The survival surface. Br. J.
Radiol. 50: 893-901, 1977.
25. Manning, MR., Cetas, T-C., Gemer, E.W.: Interstitial ther-
moradiotherapy. NCZ Monograph 61: 357-360, 1982.
26. Ovw J., Suit, H.D.: Time-temperature relationship
in hypetthermic treatment of malignant and normal tissue
in vivo. Cancer Res. 39: 3248-3253, 1979.
796 Radiation Oncology 0 Biology 0 Physics June 1984, Volume 10, Number 6

27. Rice, L.C., Urano. M., Maher, J.: The kinetics of ther-
motolerance in the mouse foot. Radial. Rex 89: 291-297,
1982.
28. Sapareto, S.A., Hopwood, L.E., Dewey, W.C.: Combined
effects of X irradiation and hyperthermia on CHO cells for
and various temperatures and orders of application. Radial.
Rex 73: 221-233, 1978.
29. Sapareto, S.A., Hopwood, L.E., Dewey, W.C., Raju, M.R.,
Gray, J.W.: Hyperthermia effects on survival and progres-
sion of CHO cells. Cancer Res. 38: 393-400, 1978.
30. Sapareto, S.A., Raaphorst, G.P., Dewey, W.C.: Cell killing
and the sequencing of hyperthermia and radiation. Int. J.
Radial. Oncol. Biol. Phys. 5: 343-347, 1979.
31 Song, C.W.: Physiological Factors in Hyperthermia of Tu-
mors. In Physical Aspects of Hyperthermia. Nussbaum, G.
(Ed.). New York, American Institute of Physics. 1982, pp.
43-62.
32. Spiro, I.J., Sapareto, S.A., Raaphorst, G.P., Dewey, W.C.:
The effect of chronic and acute heat conditioning on the
development of thermotolerance. Int. J. Radiat. Oncl. Biol.
Phys. 8: 53-58, 1982.
33. Storm, E.K., Morton, D.L., Kaiser, L.R., Harrison, W.H.,
Elliott, R.S., Weisenburger, T.H., Parker, R.G., Haskell,
C.M.: Clinical radio frequency hyperthermia: A review. NCI
Monograph 61: 343-350, 1982.
34. Tubiana, M: The future of hyperthermia. NCI Monograph
61: 539-543, 1982.
35. Westra, A., Dewey, W.C.: Variations in sensitivity to heat
during the cell cycle of Chinese hamster cells in vitro. Znt.
J. Radial. Biol. 19: 467-477, 1971.

APPENDIX I

PROGRAM TEQUIV
C
C COPYRIGHT (C) 1983, S. SAPARETO
C VERSION 2.01 LAST REVISION: 5/4/84
C
C THIS PROGRAM IS DESIGNED TO TAKE SEQUENTIAL TEMPERATURE
C VALUES AND CALCULATE THE ACCUMULATED EQUIVALENT TIME AND
C DEGREEeMINUTES CONVERTED BACK TO TIME AT A REFERENCE
C TEMPERATURE
C VARIABLES:
REAL TEMP(2000) !TEMPERATURE DATA
REAL TIME(2000) !TIME DATA
BYTE IDENT( 80) !DATA FILE IDENTIFIER
REAL TREF !REFERENCE TEMPERATURE
REAL TBREAK !BREAK TEMPERATURE (PRESET TO 43.0)
REAL TSTRT !STARTING TEMPERATURE FOR DEGREE-
MINUTE CALCULATION
C
LOGICAL NPRINT !LOGICAL VARIABLE FOR LISTING DATA
REAL RESP !TEMPORARY RESPONSE VARIABLE
REAL TSUM !EQUIVALENT TIME AT BREAK TEMPERATURE

REAL T43 !EQUIVALENT TIME AT REFERENCE

TEMPERATURE

REAL RABOVE !R VALUE ABOVE BREAK TEMPERATURE

(PRESET TO 0.5)

REAL RBELOW !R VALUE BELOW BREAK TEMPERATURE

(PRESET TO 0.25)

REAL TAVE !AVERAGE TEMPERATURE DURING DELTAT

REAL SUMRF !EQUIVALENT TIME FUNCTION
REAL DGMIN !DEGREE*MINUTES FUNCTION
REAL TDM43 !DGMIN CONVERTED TO TIME AT REFERENCE
TEMPERATURE
C
INTEGER IUNIT !TERMINAL INPUT
INTEGER OUNIT !TERMINAL OUTPUT
INTEGER LUNIT !LINE PRINTER (OPTIONAL)
 Thermal dose 0 S. A. SAPARETO AND W. C. DEWEY 797

COMMON/PRESET/ TREF,TBREAK,RABOVE,RBELOW,TSTRT

TO MODIFY DEFAULT PARAMETERS CHANGE THESE DATA STATEMENTS

BEFORE COMPILING
DATA NPRINT/.FALSE./
DATA TREF,TBREAK/43.0,43.0/
DATA RABOVE,RBELOW/O.5,0.25/
DATA TSTRT/4 I .O/
DATA IUNIT,OUNIT,LUNIT/5,7,7/
C SET REFERENCE AND BREAK TEMPERATURES
WRITE(OUNIT, lOOO)TREF
loo0 FORMAT($ENTER REFERENCE TEMPERATURE (DEFAULT=,F5.2,): )
READ(IUNIT, 10 10)RESP
1010 FORMAT(F 10.0)
IF(RESP.NE.O)TREF=RESP
WRITE(OUNIT, 1020)TBREAK
1020 FORMAT($ENTER BREAK TEMPERATURE (DEFAULT=,F5.2,): )
READ(IUNIT, 10 10)RESP
IF(RESP.NE.O)TBREAK=RESP
C READ IN DATA
CALL DATRD(TEMP,TIME,ILEN,IDENT)
C CALCULATE DOSES
T43=SUMRF(TEMP,TIME,ILEN)
TDM43=DGMIN(TEMP,TIME,ILEN)/(TREF_TSTRT)
C PRINT DATA FILE
IF( .NOT.NPRINT)GO TO 200
WRITE(LUNIT,lO5O)(TIME(I),TEMP(I),I= 1,ILEN)
1050 FORMAT(S( 1X,(,F6.2,),F5.2, ))
C OUTPUT RESULTS
200 WRITE(LUNIT, 1060)IDENT,RABOVE,TBREAK,RBELOW,TBREAK
ITREF=INT(TREF)
WRITE(LUNIT, 1070)TSTRT,ITREF,ITREF
WRITE(LUNIT, 1080)ILEN,TIME(ILEN),T43,TDM43
I060 FORMAT(/, 1X,80A 1,
/,4X;R=,F5.3,3X,FOR TEMP>,FS.l,
/,4X,R=,F5.3,3X,FOR TEMP<,FS.l,
/,lX,70(lH-))
1070 FORMAT( lX,1,6X,DATA,4X,I,4X,TOTAL,4X,(,6X,t, 10X,
1,2X,t,4X,(ABOVE ,F4.1,) 1)
/,lX,~,5X,PoI~s,3X,1,4X,TIME,5X,t,7X,I2,SX,
),3X,dm,I2,13X,I,
/,lX,70(lH-))
1080 FORMAT( lX,1,4X,I8,2X,I,F8.2, MIN I,F10.2, MIN,3X,
I,F13.2, MIN,3X,I,
/,lX,70(lH-))
STOP
END
C
FUNCTION SUMRF(TEMP,TIME,ILEN)
C
C THIS FUNCTION CALCULATES THE EQUIVALENT TIME AT THE CHOSEN
C REFERENCE TEMPERATURES.
C
REAL TEMP(2000) !TEMPERATURE ARRAY
REAL TIME(2000) !CORRESPONDING TIME ARRAY
REAL TREF !REFERENCE TEMPERATURE
798 Radiation Oncology 0 Biology 0 Physics June 1984, Volume 10, Number 6

REAL TBREAK !BREAK TEMPERATURE

REAL TSUM !EQUIVALENT TIME AT BREAK TEMPERATURE
C
REAL RABOVE !R VALUE ABOVE BREAK TEMPERATURE
C (PRESET TO 0.5)

REAL RBELOW !R VALUE BELOW BREAK TEMPERATURE

C (PRESET TO 0.25)

REAL TAVE !AVERAGE TEMPERATURE DURING DELTAT

REAL DELTAT !TIME INCREMENT BETWEEN TEMPERATURES
INTEGER ILEN !TEMPERATURE AND TIME ARRAY LENGTH
C
COMMON/PRESET/ TREF,TBREAK,RABOVE,RBELOW,TSTRT
C CALCULATE EQUIVALENT TIME
TSUM=O.O
DO 100 J=I,ILEN-1
R=RABOVE
TAVE=ABS((TEMP(J)+TEMP(J+ 1))/2.0)
DELTAT=ABS(TIME(J+ I)-TIME(J))
IF(TAVE.LE.TBREAK) R=RBELOW
TSUM=TSUM+DELTAT+R+r(TBREAK-TAVE)
100 CONTINUE
C CONVERT EQUIVALENT TIME AT BREAK TEMPERATURE TO REFERENCE TEMPERATURE
R=RABOVE
IF(TREF.LT.TBREAK)R=RBELOW
SUMRF=TSUM+R**(TREF-TBREAK)
RETURN
END
C
FUNCTION DGMIN(TEMP,TIME,ILEN)
C
C THIS FUNCTION CALCULATES THE DEGREEeMINUTES
C
REAL TEMP(2000) !TEMPERATURE ARRAY
REAL TIME(2000) !CORRESPONDING TIME ARRAY
REAL TSTRT !STARTING TEMPERATURE FOR
DEGREE-MINUTE CALCULATION
C
REAL TAVE !AVERAGE TEMPERATURE DURING DELTAT
REAL DELTAT !TIME INCREMENT BETWEEN TEMPERATURES
REAL DGMIN !ACCUMULATED TEMP*TIME ABOVE TSTRT
INTEGER ILEN !TEMPERATURE AND TIME ARRAY LENGTH
C
COMMON/PRESET/ TREF,TBREAK,RABOVE,RBELOW,TSTRT
C CALCULATE ACCUMULATED DEGREES*TIME ABOVE TSTRT
DGMIN=O.O
DO 100 J=l,ILEN-I
TAVE=ABS((TEMP(J)+TEMP(J+ 1))/2.0)
DELTAT=ABS(TIME(J+ l)-TIME(J))
IF(TAVE.GT.TSTRT)DGMIN=DGMIN+DELTAT*(TAVE-TSTRT)
100 CONTINUE
RETURN
END
C
SUBROUTINE DATRD(TEMP,TIME,ILEN,IDENT)
 Thermal dose 0 S. A. SAPARETOAND W. C. DEWEY 799

C
C THIS SUBROUTINE COLLECTS TEMPERATURE DATA FROM THE TERMINAL
C VARIABLES:
REAL TEMP(2000) !TEMPERATURE VALUES
REAL TIME(2000) !TIME VALUES
BYTE IDENT(80) !FILE IDENTIFIER
INTEGER ILEN !LENGTH OF TEMP ARRAY
INTEGER IUNIT !TERMINAL INPUT
INTEGER OUNIT !TERMINAL OUTPUT
C
C WRITTEN BY S.SAPARETO 3115183
C LAST REVISION: 3/l 5183
C
DATA IUNIT,OUNIT/5,7/
C READ THE FIRST LINE IDENTIFYING THE FILE
WRITE(OUNIT, 1000)
1000 FORMAT( ENTER IDENTIFIER (ONE LINE): )
READ(IUNIT,lOlO)IDENT
1010 FORMAT(80A 1)
C READ THE TIME AND TEMPERATURE VALUES
WRITE(OUNIT, 1020)
1020 FORMAT( ENTER TIME(MIN),TEMPERATURE VALUES, <RET>:,
+ I, (<RET> TO END DATA))
ILEN=O
DO loo 1=1,2ooo
READ(IUNIT,1030,END=2OO)TIME(I),TEMP(I)
IF(TIME(I).EQ.O. .AND. TEMP(I).EQ.O.)GO TO 200
100 CONTINUE
1030 FORMAT( 2F 10.0)
200 ILEN=I- 1
RETURN
END
C
SUBROUTINE DATRD(TEMP,TIME,ILEN,IDENT)
C
C THIS SUBROUTINE READS A TEMPERATURE DATA FILE OF THE FOLLOWING
C FORM:
C LINE 1:
C COLUMNS l-9: IDENTIFIER (9A 1)
C COLUMNS 10-13: TIME INCREMENT (14) (SECONDS)
C COLUMNS 14-80: ADDITIONAL COMMENTS
C LINE 2-END:
C TEMPERATURE VALUES SEPARATED BY A SPACE OR COMMA
C
C VARIABLES:
REAL TEMP(2000) !TEMPERATURE VALUES
REAL TIME(2000) !TIME VALUES
REALDELTAT !TIME INCREMENT BETWEEN TEMPERATURE
C VALUES

BYTE FILNAM( 14) !DATA FILE NAME

BYTE IDENT( 80) !FILE IDENTIFIER
INTEGER ILEN !LENGTH OF TEMP ARRAY
INTEGER IUNIT !TERMINAL INPUT
INTEGER OUNIT !TERMINAL OUTPUT
800 Radiation Oncology 0 Biology 0 Physics June 1984, Volume 10. Number 6

C
C WRITTEN BY SSAPARETO 912181
C LAST REVISION: 2/ 14/83
C
DATA IUNIT,OUNIT/5,7/
WRITE(OUNIT, 1000)
1000 FORMAT($ENTER FILE: )
READ (IUNIT,lOlO) FILNAM
1010 FORMAT( 14A 1)
C OPEN THE DATA FILE
OPEN(UNIT=4,NAME=FILNAM,TYPE=OLD)
C READ THE FIRST LINE IDENTIFYING THE FILE
READ(4,1020)IDENT
1020 FORMAT(8OA 1)
DELTAT=O.
C DETERMINE THE TIME INCREMENT BETWEEN POINTS FROM POSITIONS lo- 13
C OF IDENT
DECODE(80,1030,IDENT)IDELT
1030 FORMAT(9XJ4)
DELTAT=FLOAT(IDELT)/60.
WRITE(OUNIT, 1040)DELTAT*60.
1040 FORMAT($ENTER TIME INCREMENT (SECONDS)(DEFAULT=,F5.0,): )
READ( IUNIT, 105O)RESP
1050 FORMAT(FlO.0)
IF(RESP.GT.O.)DELTAT=RESP/60.
C READ THE TEMPERATURE VALUES FROM THE FILE
100 READ(4,s,END=200) (TEMP(I),I= 1,200O)
200 ILEN=I-1
C CALCULATE THE TIME VALUES
DO 300 I=l,ILEN
TIME(I)=DELTAT*FLOAT(I-1)
300 CONTINUE
CLOSE(UNIT=4)
RETURN
END